Aptos (APT)

What Aptos is

Aptos is a layer-1 blockchain built around parallel transaction execution and the Move programming language, with APT as its native asset. Aptos's documentation defines APT as the token "used for paying network and transaction fees," denominated down to an Octa, where 1 APT equals 10⁸ Octas (source: Aptos documentation, glossary, read September 2026).

Two design decisions distinguish it. The first is move language, which Aptos's documentation describes as emphasizing "access control and scarcity," so that ownership of an asset is a property the language enforces rather than a convention a contract implements. The second is Block-STM, the execution engine that runs transactions in parallel without developers having to declare in advance which state each one touches.

For where to buy Aptos from your country or state, see the availability tables on this site and the venues listed at Exchanges.

How it works

Consensus is AptosBFT, which the documentation describes as "based on Jolteon" and Byzantine fault tolerant. The network stays safe and live while up to one third of validator voting power is compromised, the standard 3f+1 arrangement for BFT consensus.

A transaction passes through five stages the documentation names explicitly: accepting, sharing, proposing, executing and consensus, and committing. Validators hold pending transactions in a mempool, and the documentation states the admission rule precisely: a validator's mempool "accepts a sequenced transaction TN only if its sequence number is greater than or equal to the sender's current sequence number." Batches are then disseminated to other validators through a Quorum Store, a leader forms a block from certified batches, the block is executed speculatively while consensus runs, and committed state is written to storage. Sequence numbers are strict at commit time: "A sequenced transaction with sequence number 5 can commit only if the account sequence number is 5."

Block-STM is what makes that pipeline fast. Aptos's documentation calls it a "state-of-the-art dynamic parallelism execution engine" that lets developers write applications "without facing design constraints of statically declaring write sets," and states it has achieved "over 160k transactions per second." The engineering idea is optimistic execution: run transactions concurrently, detect conflicts afterwards, and re-execute only what actually collided. Chains that require developers to declare read and write sets up front get parallel execution at the cost of a harder programming model; Aptos moves that cost into the runtime.

Move code arrives in two shapes. Modules are the persistent smart contracts; scripts are single-use programs that call into them. Both compile to Move bytecode executed by the Move virtual machine.

Fees have an unusual structure worth knowing. Gas unit price is denominated in Octas and set by the market above a governance floor, and higher prices buy block inclusion priority. Execution and IO costs are "burned permanently upon the execution of a transaction." Storage fees are different: they are "measured in fixed APT prices, so the permanent storage cost stays stable even as the gas unit price fluctuates," and they are refundable. The documentation describes storage fees as "burned and minted again upon refund," so freeing a storage slot returns the deposit. Paying a fixed APT price for rent-like storage means the cost of occupying state does not swing with transient congestion.

Supply and tokenomics

Aptos's technical documentation covers staking parameters in detail but does not publish an APT total supply or allocation on the pages read for this entry, so no supply figure is stated here.

Staking parameters are precise. The documentation states a minimum stake of 1 million APT and a maximum of 50 million APT, a recurring lockup of 14 days on mainnet, and mainnet epochs of 7,200 seconds, which is two hours. Rewards are paid every epoch and compound onto the principal automatically.

The reward formula is the part most often summarized incorrectly. The documentation gives it as the staked amount multiplied by the rewards rate per epoch multiplied by the number of successful proposals by the validator divided by the total number of proposals made by that validator. The rewards rate itself "is set by the Aptos governance." Crucially, only leader-validators who propose blocks earn: a validator that merely votes on other validators' proposals earns nothing for that epoch.

The staking minimum has a practical consequence. At 1 million APT, running a validator directly is out of reach for almost everyone, so most staking happens through delegation pools or exchanges, and each of those adds an operator between the holder and the protocol.

History

Aptos descends from the Diem project's engineering work, which is why Move and a BFT consensus family sit at its foundation rather than an EVM. The bet was that a language designed around resource ownership, plus an execution engine that parallelizes without developer annotations, would scale better than adding throughput to an existing design.

The 160,000 transactions per second figure in the documentation describes what Block-STM has achieved in testing rather than a sustained mainnet rate, and it is worth reading it that way. RampAtlas measures chain throughput from recent blocks over a stated window on its own chain pages rather than reproducing marketing figures.

The storage-fee refund model is a quieter but consequential choice. By pricing state occupancy in fixed APT and returning the deposit when a slot is freed, Aptos gives developers an incentive to clean up state, which is a problem most chains have handled poorly.

Risks and what to watch

Validator concentration is the first thing to check. A 1 million APT minimum stake keeps the validator set small and pushes ordinary holders into delegation, so the meaningful measure is how stake is distributed across operators rather than how many operators exist.

Liquidity of staked APT is constrained by the 14-day recurring lockup, which is a real cost if conditions change.

Client and execution-engine maturity matter on a chain built around an optimistic parallel execution model. Conflict detection and re-execution are where subtle bugs live, and a chain with a single dominant client implementation shares a single bug.

Move is a smaller ecosystem than the EVM's. Fewer audited libraries and fewer auditors fluent in the language is a real smart contract audit constraint, independent of the language's own safety properties.

Governance sets the rewards rate and the gas price floor, so both are parameters a token holder should watch rather than assume fixed.

Frequently asked questions

What is Block-STM?

Aptos's execution engine. Its documentation describes it as a dynamic parallelism engine that lets developers write applications without statically declaring write sets, running transactions concurrently and re-executing only those that conflict. The documentation states it has achieved over 160,000 transactions per second.

How long is an Aptos epoch?

7,200 seconds, or two hours, on mainnet. An epoch is the period between reconfigurations of the validator set, and staking rewards are paid at the end of each one.

What does it cost to run an Aptos validator?

The documented minimum stake is 1,000,000 APT and the maximum is 50,000,000 APT, with a recurring lockup of 14 days. Most holders therefore stake through a delegation pool or an exchange rather than running a validator.

What happens to Aptos transaction fees?

Execution and IO costs are burned permanently when a transaction executes. Storage fees are charged at a fixed APT price, and the documentation describes them as burned and re-minted on refund when a storage slot is freed.

Where can you buy Aptos?

Availability depends on your country or US state. See where to buy Aptos for the exchanges serving your jurisdiction, and Exchanges to compare fees, kyc requirements and payment methods.